Re: [h-e-w] Re: sending filenames from command-line

[Guy Gascoigne - Piggford]

One of the biggest problems with gnuclient is that it's been hacked
about by a number of people over the years and as such I think that it's
tricky to correctly get it's copyright handed over the FSF, and unless
I'm mis-remembering, this is a requirement for it to be included as part
of emacs.

Secondly emacsclient only used unix domain sockets and doesn't have any
cross machine support.  More to the point every time this comes up on
the emacs list there are some quite reasonable security concerns about
adding such to emacsclient. Needless to say, gnuclient has cross machine
support and personally I'm happy to accept the minimal security that it
exhibits.

There has been talk about adding tcp/ip socket support to emacsclient
(heck, I've been one of the ones talking), but no one has actually
stepped up to do it.  Personally I find that the old gnuclient mechanism
is a much simpler structure to hack on since it can pretty much be
worked on in isolation from emacs itself.  Emacsclient requires much
more support from emacs itself, you need to be comfortable changing and
building emacs, as well as lisp as well as the emacsclient C code.  And
this is why I've not fixed it myself, I'm happy hack gnuclient and I've
not had the time to hack emacsclient yet - they are just not that similar.

Guy

address@hidden wrote:
> Sounds reasonable.  The emacsclient in the CVS version doesn't 
> seem to have much functionality, compared to gnuclient. 
> What's the benefit of emacsclient over gnuclient? That it's 
> officially part of GNU Emacs? 
> What's the problem with emacsclient on Windows anyway? 
> Doesn't Windows support Berkeley sockets?
>
> Rob
-- 
Guy Gascoigne-Piggford
www.wyrdrune.com